Optimal starting point for antiretroviral HIV treatment in a town in Cameroon: a randomised controlled study

With the 2010 and 2013 guidelines, WHO recommended earlier initiation of antiretroviral treatment (ART) than in previous guidelines [1]. Evidence for a beneficial effect of earlier start of treatment was shown first in Western countries [2‐4]. Later on, studies from African countries and Haiti have supported this [5‐8]. Population effects pointed in the same direction [9‐11]. Randomised studies are few and there has recently been a call for more research about when to start ART in Africa [12]. One reason we started our study in 2007, was precisely because our clinical impression from around 2003 was that ART was initiated too late. Other authors had argued that WHO patients with clinical stage 3 and 4 should initiate treatment [13]. ART became accessible in Cameroon around 2002-4, both in terms of drug supply and through a sharp decline in prices, ending with free distribution of some drugs in large parts of the country from around 2005. CD4 counts have been accessible more or less continuously from 2005, but are usually free of charge only during campaigns and for pregnant women. Recommendations for low income settings before 2010 were to initiate treatment when a patient’s CD4 count dropped below 200 cells/mm³. National guidelines were in agreement with this. We report results from our study, which to some extent answered the original question of when to start, but also revealed some unexpected experiences.

All new cases of adult treatment-naïve HIV-positive patients diagnosed between 15 March 2007 and 31 December 2008 at the Protestant Hospital in Ngaoundere, a town of 300 000 inhabitants in Northern Cameroon, were considered eligible for the study. They were randomised to initiating treatment with CD4 counts of either <250 or <350 cells/mm³, or when their condition after randomisation progressed to WHO clinical stage 3 or 4. Exclusion criteria were:

Enrolled patients who did not start treatment right after randomisation were scheduled to have clinical and CD4 controls free of charge every three months. After start of treatment, control of CD4 was offered every six months. Outcome measures were WHO clinical stage, mortality, changes in CD4 and weight maintenance. Because patients were to be recruited in WHO stage I-II, we did not expect many deaths during the three and a half years our study was meant to last. Power calculations therefore were made based on substitute outcomes, i.e. differences in cd4 values and weight loss. A mean group difference in final CD4 value of 100 cells/mm³ and a mean weight loss difference of two kilos were considered clinically important. In our calculations we used anticipated cd4 values of 400 and 300 cells/mm³ and a standard deviation of 100, and for weight loss 5 ± 2 kg and 3 ± 1.5 kg. To show such differences, calculations with significance level 5% and a statistical power 90% in both cases suggested that 21 patients in each group were needed [14].

The choice of drugs for ART in Ngaoundere was very limited. In practice almost all patients in 2007-8 received a one drug combination of lamivudine, stavudine and nevirapine. The exception was patients previously treated for Tb, who received efivarenz instead of nevirapine; three patients in our study. From 2008-9 serious stavudine side effects caused a gradual change of the three-drug pill to one where stavudine was replaced by zidovudine. This was the same for both groups. Cotrimoxazole was given routinely to all patients receiving ART, and also to other patients coming for routine follow-up.

Patients who started ART were informed about the reason for this, either cd4 level, WHO clinical stage, or both.

The study was scheduled to follow patients at least until the end of 2010. We terminated our study in July 2010, six months before planned, because both the WHO and the Cameroonian recommendations had changed during the preceding months, and the changes were in accordance with our early start of treatment. We then actively searched for the deferred therapy patients who did not present themselves according to their follow-up plan. Postal street addresses do not exist in Ngaoundere, but some of the patients could be contacted by mobile phone. During the first months of 2011 we were able to carry out the planned independent clinical evaluation for half of the original patients. It took some time to find the whereabouts of all patients, and censoring as dead, lost-to-follow-up or having been clinically evaluated was done with 30 June 2011 as the ultimate date.

Between April 2007 and November 2008, 424 HIV positive patients consulted at the Protestant Hospital. Most patients fulfilled one or more of the exclusion criteria. They therefore did not go through the randomisation procedure, and most of them started ART within a few weeks. CD4 was measured in 381 patients. Mean values were 174 cells/mm³ for 227 females and 138 cells/mm³ for 154 males. Forty-four patients were enrolled in the study, 28 females and 16 males, with a mean age of 31 and 36 years, respectively. Five patients were pregnant when they were diagnosed and recruited into the study, they did not previously know they had seroconverted and were ART-naïve like the other patients. Baseline characteristics are shown in Table 1. Exclusions are shown in Figure 1. During follow-up, patients had from zero to seven visits with CD4 count, more regularly after initiated ART therapy because this usually meant monthly visits with renewal of drug supply.

Fourteen patients randomised to early treatment started ART, 13 because of CD4 < 350 cells/mm³, the last patient erroneously because an attack of herpes zoster was considered as stage 3. This patient later died. Eleven patients randomized to deferred treatment started ART, seven of them because they went below the threshold of <250 cells/mm³, four because they reached WHO stage 3. Two females randomised to early treatment, and four females and two males randomised to deferred treatment, died. Of the eight dead patients, one ‘early’ and three ’deferred’ patients had not started ART when they died 3 to13 months after randomisation. The other four died from 9-39 months after randomisation and 3-39 months after ART had been started. Figure 2 suggests an improved survival in the early treatment group, although not significant in such a small sample (HR = 0.26, 95% CI 0.05-1.29). Twenty-two patients went through the planned clinical evaluation at the end of the trial. One patient in the deferred group had AIDS, one in the early group was in WHO stage 2, the twenty others were asymptomatic (Table 2). Mean CD4 in the ‘early’ group was 556 and in the ‘deferred’ group 382 cells/mm³, this difference is not significant.

Of the 14 patients lost to follow-up, three patients randomised to early treatment and one to deferred treatment moved to other towns, but some had occasional telephone contact with the hospital. Patients from distant locations were offered travel money if they would come to the final evaluation, but only one came. Vain efforts were also made to have patients who had moved inform us of their CD4 values taken elsewhere. Six of the eight dead patients never had any CD4 follow-up, and so was the case for three "early" and four "deferred" patients lost to follow-up. The average BMI for the evaluated patients was similar for the two groups. Of the dead patients, two had no schooling, four had 1-4 years and two had 5-8 years of schooling.

Our study had a similar design as the Haitian study by Severe et al [6], who started their study in 2005 and had AIDS as their clinical exclusion criterion. When our study started two years later, the CD4 criteria in Cameroon followed WHO and national guidelines, with ART-start when CD4 dropped below 200 cells/mm³. However, it had become common to start treatment for patients in WHO clinical stage III. At that time in Cameroon, it therefore seemed unethical to expose stage III patients to deferred treatment, and it turned out that a majority of potential study patients had to be excluded because of this. With few patients and incomplete follow-up, our results are much less clear than the Haiti study, but their tendency is the same. Deferred treatment seemed to carry no increased survival or other clinical advantage. Sensitivity analyses were performed where all patients lost to follow-up and not known to be alive were counted as dead, and with all patients lost to follow-up counted as alive, but these analyses did not change conclusions (HR 0.42 (95% CI 0.15-1.22), and HR 0.33 (95% CI 0.07-1.66)), respectively. Counting the evaluated patient with AIDS among the dead gave only minimal changes. With few studies adding to current evidence, we think our study tells something about existing practical possibilities of HIV therapy and follow-up, which are probably common to many places in Africa.

It is not surprising that a majority of the patients who came for evaluation were among the least sick, but it is striking and encouraging that almost all were asymptomatic. This is in agreement with our clinical impression that patients who come regularly for their ART drugs in most cases improve clinically and remain well. Also, patients coming regularly to controls and who have a relatively high CD4 value usually avoid serious clinical deterioration before ART can be started. This is in spite of the limited local therapeutic offer of one standard ART combination and one available second line alternative. In Cameroon, scale-up of ART has been shown to be associated with increased drug resistance mutations [15], and resistance monitoring may be the next priority after CD4 access [16]. Routine measuring of viral load seems to be a lesser priority [17, 18]. In 2013, consolidated WHO guidelines have recommended early start of ART on an even broader base [1].

Median CD4 and BMI were slightly lower at baseline for the deferred group, which had relatively more females. The difference may have contributed to the higher number of dead or lost to follow-up, but we think the deferred treatment was more important. It may be worth noting that all the dead patients had little or no schooling.

To get an idea about natural CD4 decline without treatment, and thus about the lead time bias for the deferred treatment patients, we had foreseen an auxiliary study of patients who had at least one repeat measure of CD4 before ART became common. For this purpose, we went through hand-written medical records for all patients during the years 2003-7 at an HIV clinic in the region. We found 154 patients with repeat measures before any written signs of prescriptions of ART drugs. However, many of these patients had unexplained increases of CD4, which suggested that they had been treated in spite of no mention of this in the medical record, and we have not used these data. Without such data and also missing CD4 data for several patients, differences in mean CD4 values in the two groups are difficult to interpret. In the Ivory Coast it has been shown that 75% of ART-naïve patients had values of less than 350 cells/mm³ within seven years after seroconversion [19].

The study was cleared by The Regional Committee for Medical Research Ethics in Western Norway (REK III nr 115.02), and permission was given by the Provincial Head of medical services in Adamawa province where Ngaoundere is located. The study was reported to the Norwegian Social Science data services, project no 9248/02. The study was web-registered at ISRCTN22114173 http://​www.​controlled-trials.​com/​isrctn/​search.​html?​srch=​22114173&​sort=​3&​dir=​desc&​max=​10&​Submit=​SUBMIT. The CONSORT checklist has been used for the abstract.

Funding was modest, approximately 9-10000 Euro a year between 2006 and 2012, mainly used to pay the laboratory technician who has been a stable clinical first line health worker throughout the study. Part of the money covered the scheduled follow-up consultations, including local transport and blood tests, sometimes also for non-ART drugs not covered otherwise. Travel between Cameroon and Norway has been covered for the author KH.